Skip to content
All projects

DevOps case study

Cloud Deployment Dashboard

Deployment visibility across environments with build status, rollbacks and environment variable auditing.

Rollback time
30m → 90s
Failed deploys caught
pre-prod
Env drift incidents
→ 0
zsh — deploy
 vercel deploy --prod
Building… done in 24s
Uploading assets… ok
 Production: live
Rollback target: a91f3c2

Interface reference — rendered from the real component structure, not a stock image.

The problem

Nobody could say which commit was live in which environment, and rollbacks meant a frantic search.

The solution

One dashboard aggregating build status per environment, one-click rollback to a previous deployment, and change history for environment variables.

Hard parts

Keeping secrets out of the UI while still showing what changed — the dashboard shows keys and timestamps, never values.

Outcome

Recovery from a bad release dropped from a stressful half hour to under two minutes.

Technology used

  • Docker
  • GitHub Actions
  • Vercel
  • Node.js
  • Next.js

How it was delivered

  1. 01Client problemDeployment state lived in three dashboards and one person's memory.
  2. 02ResearchDocumented the existing release process and its failure points.
  3. 03PlanningAggregate rather than replace — the dashboard reads existing platforms.
  4. 04WireframeEnvironment columns, commit rows, one rollback action.
  5. 05DevelopmentGitHub and Vercel APIs polled and cached, audit log written on every action.
  6. 06TestingRehearsed rollbacks against a staging project until it was boring.
  7. 07DeploymentRead-only first, write actions enabled after a week of observation.
  8. 08ResultsReleases stopped being an event.

Lessons learned

Show keys and timestamps, never secret values. Auditability and exposure are not the same thing.

More case studies